Join us this afternoon in the Greenberg Room as we hear a colloquium from Jacob Eisenstein (Georgia Tech) today at 3:30pm. A department social will follow.
Variation and Change in Online Writing
Online writing is an increasingly ubiquitous mode for informal, phatic communication, but the implications of this shift for the relationship between writing and speech are still contested. While some point to the rise of a new “netspeak” dialect, quantitative analysis suggests a picture that is more complex: online writing reproduces lexical and phonetic variation from spoken language, while simultaneously hosting an impressive array of apparently novel orthographic variables. I will present computational statistical techniques for identifying variation in online writing, and will discuss the geographical, social, and linguistic properties of several types of variables. Next, I will consider the diachronic perspective, where large-scale longitudinal data enable robust inferences about patterns of linguistic influence across thousands of lexical variables. I conclude with ongoing work on whether and how orthographic variables can maintain their social and geographical distinctiveness in the face of pressure towards leveling in an ever more densely-connected online world.
The department welcomes Bill Croft (University of New Mexico), who will be speaking at 3:30pm today, February 6 in the Greenberg room. All are welcome, with a social to follow. The title and abstract are given below.
Force-dynamic image schemas: between verb and argument structure construction
There has been a long debate over the relative semantic contribution of verbs and argument structure constructions to the structure of the event expressed by a clause, in particular semantic structures described as “transfer”, “emission”, “application” and so on. In this talk, I will argue that these semantic structures are force-dynamic image schemas that are only partly constrained by verb meaning and argument structure construction. Verbs have a force-dynamic potential that allows them to be construed in more than one force-dynamic image schema. Argument structure constructions constrain but do not determine the force-dynamic structure of the event expressed by the verb occurring in the construction. The mapping between verb roots and force-dynamic image schemas varies across languages, but additional research is required to determine what constraints there are on possible mappings.
Roger Levy (UC San Diego) will give a joint Linguistics/Psychology colloquium Wednesday 1/28 at 3:45PM in Jordan Hall Room 041 (420-041). There will be a reception in the Jordan Hall Lounge immediately afterward.
Expectation-based language comprehension and production
Using language to communicate is central to what makes us human. Elucidating the knowledge, expectations, and cognitive resources that allow us to communicate so effectively is one of the most fundamental problems in the study of mind. For much of the contemporary history of psychology and linguistics, motivated by the ideas of figures including Chomsky, Miller, and Fodor, work on this problem has conceptualized language processing as centrally about modular structure-building operations and the memory resources required to carry them out. Here I describe an alternative approach that conceptualizes language processing as centrally about rational, goal-driven inference and action. First, I outline a state-of-the-art theory of expectation-based incremental language understanding, in which comprehenders integrate diverse information sources from preceding context to guide interpretation of current input. This theory unifies three key, seemingly disparate topics in the domain of language understanding — ambiguity resolution, prediction, and syntactic complexity effects — and finds broad empirical support in data from both controlled experiments and naturalistic language understanding. Second, I describe several apparent empirical puzzles for this theory that ultimately lead us to revisit one of the implicit foundational assumptions in all theories of language understanding: that of modularity between the processes of word recognition and of inter-word, utterance-level comprehension. I generalize the expectation-based theory to a fully bidirectional, interactive theory of word recognition and utterance comprehension, and show how this generalized theory solves the apparent puzzles and leads to a range of new, empirically verified predictions. Finally, I touch briefly on the consequences of this view for language production: why do speakers choose to structure their utterances the way they do? The expectation-based theory novelly predicts that speakers use the options afforded them by their native language to come as close as possible to a uniform distribution of information content throughout their utterance. We confirm this prediction through statistical analysis of speaker choices regarding optional word omission in naturalistic speech.
Our next departmental colloquium will be given by Paola Merlo (University of Geneva), on Friday 1/23 at 3:30 in the Greenberg Room.
Linguistic issues in Multi-lingual Natural Language Processing
Current natural language technology shows great interest in multi-lingual tasks. Multi-lingual approaches leverage two apparently contradictory properties of all languages: languages are both very different from each other in their lexical and grammatical properties, but they are very similar to each other at more abstract levels. While
the goal of these multi-lingual tasks is driven by technology, there are linguistic lessons to be gleaned by looking at what features, linguistic properties, annotations and tools are valid across languages, allowing us to begin to test theories of language universals on a truly large scale.
I will discuss two case studies related to this topic. The first case study shows that corpus data and typological data involving the causative alternation exhibit interesting correlations explained by the notion of spontaneity of an event. This notion also explains patterns of variation in the translation of the causative construction
in parallel corpora. The second case study is work in progress: we use counts collected from tree-annotated dependency corpora to tease apart different theories of language universals.
Work in collaboration with Tanja Samardzic and Kristina Gulordava.
Sarah Murray (Cornell University) will be giving a colloquium next Friday at 3:30 PM in the Greenberg Room. Join us as we get the 2015 colloquium series off to a strong start! All are welcome.
Complex connectives: Building disjunction from conjunction
Abstract: This talk discusses the interpretation and analysis of several connectives found in Cheyenne (Algonquian), drawing on the author’s fieldwork as well as several collections of texts. Coordinating connectives in English, including and (conjunction), but (contrastive conjunction), and or (disjunction), are monomorphemic. In Cheyenne, the basic form used for conjunction is naa. Other connectives are morphologically complex, formed by combining naa with other morphemes, all of which have independent uses.
These complex connectives, and certain uses of naa alone, complicate a compositional, truth-functional analysis of the Cheyenne connectives. In particular, though disjunction is logically weaker than conjunction, the two forms for disjunction – naa matȯ=héva and naa mó=héá’e – each contain the conjunction naa. Recently, several authors have proposed analyses of related data from other languages where the basic element, similar to naa, is not a true conjunction. However, the data differ from Cheyenne in crucial ways. Building on these analyses, and other recent proposals on the semantics of disjunction, this talk proposes an analysis of the Cheyenne connectives that preserves naa as conjunction.
Embedded implicatures as pragmatic inferences under compositional lexical uncertainty
Christopher Potts, Stanford Linguistics
Monday, January 12, 2015
Building 460, Room 126 (Margaret Jacks Hall)
How do comprehenders reason about pragmatically ambiguous scalar terms like ‘some’ in complex syntactic contexts? In Gricean theories of conversational implicature, local exhaustification of such terms (‘only some’) is predicted to be impossible if it does not strengthen the literal meaning, whereas grammatical accounts of implicature predict such construals to be available. Recent experimental evidence supports the availability of these local enrichments, but the grammatical theories that this evidence supports do not provide viable mechanisms for weighting such construals against others. We propose a probabilistic model that combines previous work on pragmatic inference under ‘lexical uncertainty’ with a more detailed model of compositional semantics. We show that this model makes accurate predictions about new experimental data on embedded implicatures in both non-monotonic and downward-entailing semantic contexts. In addition, the model’s predictions can be improved by the incorporation of neo-Gricean hypotheses about lexical alternatives. This work thus contributes to a synthesis of grammatical and probabilistic views on pragmatic inference.
This talk represents joint work with Daniel Lassiter, Roger Levy, and Michael C. Frank.
Join us in the Greenberg Room today at 3:30 PM for a colloquium talk by Teenie Matlock (UC Merced).
Aspect and Metaphor in Framing
Framing plays an important role in everyday communication and reasoning. People constantly frame events, states, and situations with the goal of persuading others to form particular attitudes and take particular actions. This is well known across the social sciences. Still, little is known about the role of linguistic semantics in this process, especially when it comes to subtle shifts in and literal and non-literal meaning. This presentation will review cognitive linguistics research on framing and discuss recent experimental findings on how subtle shifts in aspectual and metaphorical information (e.g., manner of motion in non-literal verbs) can result in different inferences in the interpretation of political messages.
Our next colloquium speaker will be Kathryn Campbell-Kibler (Ohio State), at 3:30PM today, Oct 31 in the Greenberg room.
Reducing sociolinguistic cognition to previously unsolved problems
More than half a century of research in language variation and related fields has documented speakers’ ability to alter small details of their speech to conform to or agentively change the social elements of an interaction. Likewise, listeners are able to note these speech patterns and use them to form or change their social reading of a situation. These abilities apply both to linguistic forms speakers can verbally describe or even manipulate on command, and to those they cannot. In this talk I discuss the cognitive structures necessary to accomplish these feats. I consider the history of the sociolinguistic monitor, variation’s most developed model, and discuss its shortcomings in light of current evidence. I propose that sociolinguistic cognition requires no specialized cognitive machinery, rather its patterns are explainable by independently motivated structures of linguistic and social cognition and their interactions. Given that linguistic and social processing are both as yet not fully understood, the study of sociolinguistic cognition can help illuminate their structure by examining their interactions. To do so successfully, both the linguistic and the social must be centered.